The UZH Science Cluster server went into production mode in August 2025. Thanks to a broad set of improvements and testing during the beta phase, the production server is now more accessible, more performant and more stable. Stream your OME-Zarrs in the Vizarr web viewer and verify segmentation quality right in the browser. Interactively train a classifier in napari by streaming the data remotely over https. Or evaluate your measurements in the new Fractal feature explorer web dashboard to check if your experiment has worked.

<aside> 💡

Fractal server is now available at https://fractal-scluster.mls.uzh.ch

The old Fractal URL will forward to it.

</aside>

Image streaming options

Using the new fractal-data service, large images stored remotely are now accessible to be streamed into both the web viewer as well as into the napari viewer.

Vizarr viewer streaming:

https://www.youtube.com/watch?v=6jjZ264z8ZM

napari viewer streaming:

https://www.youtube.com/watch?v=eAzm10Vrhu0

Loading remote OME-Zarrs requires napari with the napari-ome-zarr-navigator plugin (version 0.2.0 or newer). Other plugins shown here are the napari-feature-classifier & the napari-feature-visualization.

Fractal feature explorer

You can explore the feature measurements interactively in the Fractal feature explorer web dashboard. It streams feature measurements from the OME-Zarrs and allows for interactive filtering, overview plots as well as data export.

https://youtu.be/XOwIGbjdktg

Flexibility

Thanks to the Fractal flexibility update, workflows in Fractal are much more granular and interactive: Try your workflow on a single image, or apply it to thousands of images at once, all just one click away. Our improved history functionality keeps track of all your runs and knows which images were processed successfully.

https://www.youtube.com/watch?v=Jdr7VOiv7k0

Improved performance & stability

We have optimised the overheads of workflow submission and improved the stability of the connection to the Science Cluster. Even with many users submitting workflows at once, your workflow will keep running now.

Thanks to a multi-level backup approach, your projects, datasets and workflows are now safer than ever.